Better Way to Aggregate and Assign Random Winner - node.js

Question

I'm trying to aggregate a set of transactions using the data set below and choose a winner in every grade. The winner is randomly chosen from within the grade.
{ "_id" : ObjectId("5ce6fb4b3d1be918e574500a"),
"eventId" : ObjectId("5ce2f540bf126322a6be559b"),
"donationAmt" : 32,
"ccTranId" : "HzP4B",
"firstName" : "Jason",
"lastName" : "Jones",
"grade" : "1",
"teacher" : "Smith, Bob",
"studentId" : 100 },
{ "_id" : ObjectId("5ce6fb4b3d1be918e574500b"),
"eventId" : ObjectId("5ce2f540bf126322a6be559b"),
"donationAmt" : 15,
"ccTranId" : "HzP4A",
"firstName" : "Joey",
"lastName" : "Jones",
"grade" : "1",
"teacher" : "Smith, Jane",
"studentId" : 200 },
{ "_id" : ObjectId("5ce6fb4b3d1be918e574500c"),
"eventId" : ObjectId("5ce2f540bf126322a6be559b"),
"donationAmt" : 25,
"ccTranId" : "HzP4D",
"firstName" : "Carrie",
"lastName" : "Jones",
"grade" : "2",
"teacher" : "Smith, Sally",
"studentId" : 300 }
I'm using this script to aggregate.
Donation.aggregate([
{
$match: {
eventId: mongoose.Types.ObjectId(eventId)
}
},
{
"$group": {
"_id": "$studentId",
"first": { "$first": "$firstName" },
"last": { "$first": "$lastName" },
"grade": { "$first": "$grade" },
"teacher": { "$first": "$teacher" }
}
},
{
"$group": {
"_id": "$grade",
"students": {
$push: '$$ROOT'
}
}
}
, { $sort: { _id: 1 } }
])
The output gives me this to work with. Then, I iterate through the each element and assign one of the students in the subdocument as winner.
The double group seems sloppy and it would be nice to execute an expression within a $project clause to randomly assign the winner.
Is there a cleaner way?
{
"_id":"1",
"students":[
{
"_id":100,
"first":"Jason",
"last":"Jones",
"grade":"1",
"teacher":"Smith, Bob"
},
{
"_id":200,
"first":"Joey",
"last":"Jones",
"grade":"1",
"teacher":"Smith, Jae"
}
]
},
{
"_id":"2",
students":[ .... ]
},

Answer 1

Random means that you need to get unpredictable results. The only operator that can help you in MongoDB is $sample. Unfortunately you can't sample arrays. All you can do is to apply filtering condition and then run { $sample: { size: 1 } } on that filtered data set:
db.col.aggregate([
{
$match: {
eventId: ObjectId("5ce2f540bf126322a6be559b"),
grade: "2"
}
},
{ $sample: { size: 1 } }
])
To make it a little bit more useful you can take advantage of $facet and run multiple samples for every grade in one query:
db.col.aggregate([
{
$match: {
eventId: ObjectId("5ce2f540bf126322a6be559b")
}
},
{
$facet: {
winner1: [
{ $match: { grade: "1" } },
{ $sample: { size: 1 } }
],
winner2: [
{ $match: { grade: "2" } },
{ $sample: { size: 1 } }
]
// other grades ...
}
}
])